Nonlinear reduced order homogenization of materials including cohesive interfaces
Nonlinear reduced order homogenization of materials including cohesive interfaces
复制标题
DOI：
10.1007/s00466-015-1163-0
复制
发表时间：
2015-07
影响因子：
4.1
通讯作者：
F. Fritzen;Matthias Leuschner
中科院分区：
文献类型：
--
作者：
F. Fritzen;Matthias Leuschner
复合材料的力学响应强烈地受到组分之间界面的非线性行为的影响。为了对这些材料进行可靠但计算效率高的预测，开发了一种降阶模型。NTFA（Michel和Suquet，Int J Solids Struct 40：6937-6955，2003，COMPut Methods Appl Mech Eng 193：5477-5502，2004）和pRBMOR（Fritzen，Hodapp和Leuschner COMPut Methods Appl Mech Eng 260：143-154，2013，Fritzen等人，Comput Methods Appl Mech Eng 278：186-217，2014）。其核心思想是通过一个整体的降阶基函数来参数化粘性界面上的位移跳跃。微观力学的考虑和潜在的本构模型的结构导致变分制剂和减少平衡条件。预分析阶段的准确性的影响进行了研究，使用几何最优的训练方向。减少模型进行测试的三维微观结构。除了有效应力响应，拉压不对称性和分离界面的分布进行了研究。实现了大约的内存节省。计算时间大大减少。
The mechanical response of composite materials is strongly influenced by the nonlinear behavior of the interface between the constituents. In order to make reliable yet computationally efficient predictions for such materials, a reduced order model is developed. Conceptual ideas of the NTFA (Michel and Suquet, Int J Solids Struct 40:6937–6955, 2003, Comput Methods Appl Mech Eng 193:5477–5502, 2004) and of the pRBMOR (Fritzen, Hodapp and Leuschner Comput Methods Appl Mech Eng 260:143–154, 2013, Fritzen et al., Comput Methods Appl Mech Eng 278:186–217, 2014) are adopted. The key idea is to parameterize the displacement jumps on the cohesive interfaces by a reduced basis of global ansatz functions. Micromechanical considerations and the potential structure of the constitutive models lead to a variational formulation and reduced equilibrium conditions. The effect of the preanalysis phase on the accuracy is investigated using geometrically optimal training directions. The reduced model is tested for three-dimensional microstructures. Besides the effective stress response, the tension–compression asymmetry and the distribution of the separation of the interface are investigated. Memory savings on the order ofare realized. The computing time is reduced considerably.